THE RELATIONSHIP OF LOCAL PUBLIC EXPENDITURES AND RESIDENTIAL PROPERTY VALUES IN MASSACHUSETTES TOWNS by John Benjamin Fleeman A. B., Bowdoin College, Economics and Government (1978) Master of Science in Management, Florida International University (1981) Submitted to the Department of Architecture in Partial Fulfillment of the Requirements for the Degree of MASTER OF SCIENCE in Real Estate Development at the Massachusetts Institute of Technology September, 1990 (c) John Benjamin Fleeman The author hereby grants to MIT permission to reproduce and to distribute copies of this thesis document in whole or in part. Signature of Author 7 of Archicur July 30, 1990 '~IM~artment Certified by Mard A. Louargaind Lecturer, Department of Urban Studies Thesis Supervisor Accepted by______________ Glori'19Shuck Chairperson Interdepartmental Degree Program in Real Estate Development MASSACHUSETTS INSTITUTE OF TECHNPr Afy SEP 19 1990 LIBRARIES THE RELATIONSHIP OF LOCAL PUBLIC EXPENDITURES AND RESIDENTIAL PROPERTY VALUES IN MASSACHUSETTS TOWNS by John Benjamin Fleeman Submitted to the Department of Architecture on July 30, 1990 in partial fulfillment of the requirements for the degree of Master of Science in Real Estate Development ABSTRACT This is a study of the relationship between local public expenditures and single family residential property values in several Massachusetts towns. The majority of the literature regarding the general relationship of public expenditures and property values indicates that the value of public services is capitalized into higher home prices, while higher property tax rates are capitalized into lower home prices. Data regarding actual home sales and public expenditures in ten towns are used in a multiple regression model to test this hypothesis. The results indicate a positive relationship between home values and police and education expenditures, but also, a positive relationship between home values and tax rates. Closer scrutiny reveals that other factors positively correlated with both tax rates and home values, such as restrictive zoning and per capita income, lead to a positive relationship between tax rates and home values, contrary to the expectations of an "all else equal" model. Thesis supervisor: Marc A. Louargand, Ph.D. Lecturer Department of Urban Studies and Planning ACKNOWLEDGEMENTS I would like to thank my wife, Diana, for the love, support and encouragement she has given me not only during the production of this thesis, but also throughout more than six years of marriage. Thanks also go to our daughter, Rebecca, who brought both of us great joy during this first year of her life. Finally, to my advisor, Marc Louargand, for his patient and positive guidance through the thesis process. John B.Fleeman Cambridge,MA July, 1990 TABLE OF CONTENTS I ) INTRODUCTION....................................... 5 II) LITERATURE REVIEW...................................12 III) THE MODEL..........................................29 IV) RESULTS, INTERPRETATION AND DISCUSSION.............41 V) CONCLUSIONS........................................ 48 VI) APPENDIX I......................................... 51 VII) APPENDIX II........................................60 VII) REFERENCES......................................... 62 LIST OF TABLES 1. Summary of Expected Signs............................41 2. Regression Results...................................42 I) INTRODUCTION This is a study of the relationship between local single family residential property public expenditures and values in several Massachusetts towns. of number studies the provided services reflect that study draws relationship, past which and governments local for homes provide more public services. in few variables a have hypothesis that people previous work from and adds by have been a paying higher prices valuation by located in communities which This the in confirm or reject the attempted to value conducted There the analyzing not observed in public services in previous studies. reflection of The higher home such The considering a provides a consumers among housing indifferent a relatively Further, assume that stock both difference between Suppose are other is the availability must be a family is A, which low level of public all aspects of the housing identical living in of services, and level of public relatively high towns housing choosing between Town services. in a condition in of choices. move and is which provides capitalization of to as a is that concept equilibrium, economic value of prices is referred services. Town B, the so one town that the as opposed of public services. only to the Given those circumstances, it is clear that the family will choose Town A and enjoy the greater availability of incurring any additional cost. services without On a broader scale, this will prospective homebuyer choosing Eventually, in the absence sites, everybody the consumer is indifferent between the towns Thus, public services additional available increases in At what stop? process up until housing in Town housing in Town B that between living in lower priced and higher priced housing in the housing reflect how will towns. However, scarcity bidding-up housing in low-service Town B A. A. every these two for housing enough relative to high-service Town of scarcity of a Theoretically, prices will be bid A is expensive true housing will be bid up. this will level of as the demand Town A, the price of such price between would live in Town factor, and is a be consumers value available in value of the additional services price difference Town A, the and the is said to be capitalized into the housing prices in Town A. The problem attempting of to demonstrate capitalization with empirical data is that far more factors than just prices. the provision of public The differences, such theory but in described above reality, it uniformity among the purchase with issues services affect housing relating to is difficult housing choices. of a home, consumers the site, their place(s) of employment assumes away such to imagine When considering typically are concerned such as distance from and neighborhood security, as well as issues relating to the dwelling itself, such as the size of the house, its age, and style. explain the relationship of In attempting to housing prices to local public expenditures, a factors study must so that hold constant for they will not bias these other the estimate of how public services relate to property values. The most common approach to this problem is the use of hedonic pricing concept that the models. Such commodity under case, housing) is really (more accurately, a are based on consideration (in this The market price of a house set of housing services dwelling) the a collection of commodities, each of which has a distinct value. particular models is the sum of provided by a the prices individual commodities (see Goodman [2], of the Griliches [3], and Sullivan [15]). By collecting data regarding these various components housing of statistical services technique of and the multiple linear use of the regression, the influence of each commodity on the overall price of housing services can be estimated*. For example, if we were vary solely with square to assume that housing prices footage, distance to Business District (CBD) and the collect data regarding the Central age of the house, we would those independent variables and estimate a linear relationship of the form: P = a + bX 1 - cX 2 - dX 3 where P is the price of a house and a is the intercept (the value of P when the independent variables are zero), X1 is square footage, X 2 is distance from the CBD, X 3 is the age ------------------------------------------------------ * - It is assumed that the reader has a basic knowledge of multiple linear regression and the interpretation statistical significance of equation coefficients. of the of the house, and b, c and d are the coefficients of X, X 2 and X3 , respectively. Each coefficient relates the expected change in independent P for a one unit change variable, assuming variables in the equation sign for and price. square footage results negative signs for c other are held constant. b implies a positive footage (Xj) all That in an increase by b increase in in price. The and d indicate negative relationships price. unit decreases by c, The positive is, a one unit from the CBD (X2 ) and a independent relationship between square between both distance For in its related one increase in age (X3) with distance, and for a one unit increase price in age, price decreases by d. Hedonic pricing models estimate prices for a collection of utilize regression analysis to heterogeneous commodities comprised of distinct commodities. The independent variables are the distinct commodities and the coefficients of each variable distinct are the indicators of how a commodity commodity. Thus, magnitude and affects the the models price attempt direction of each of to unit of that the overall show both variable by the the size and sign of the coefficient. This study is concerned relationship between single-family residential local hedonic pricing model which local public expenditures with public property explaining the expenditures and values, and uses a includes several categories of as independent variables. independent variables thought to Other have a significant impact upon single-family variable) are linear residential values analysis is used dependent and then multiple included in the database, regression (the to determine what relationship exists. From a standard hypothesis (HO) statistics is that local expenditures single-family residential upon coefficients perspective, for the public statistically significantly property the null have no effect values. expenditure If the variables are different from zero, the null hypothesis is rejected. Taking a less rigorous that there is a approach, the expectation is positive relationship between local public expenditures and single-family residential property values. As discussed in our initial example seems reasonable to expect that pay more for more than houses if the houses were B, it people would be willing to houses located in a public services of Towns A and community which provides they would pay for the same located in a community providing fewer public services. Another factor effect of property in the capitalization process taxes. Lower property is the taxes are presumed to be capitalized, yielding higher housing prices. If a homebuyer is faced with a choice between two identical homes in the property tax same location which burden, the differ buyer should only be willing in the to pay more for the home for which taxes are lower. However, property taxes are usually the main source of revenue for local governments. Thus, in order to provide public services, property taxes. local governments must be To the extent demanded, it is likely that Thus, one might expect that able to impose more services are property taxes will be higher. to observe opposing forces of capitalization - a negative relationship between prices and tax rates and a positive relationship between public expenditures. The net result will prices and depend upon how homebuyers view the benefits of public services relative to the cost of the greater than If not, taxes. the costs, viewed as the perceived benefits housing prices should the net result taxes are If be higher. will be lower housing a direct are prices. payment, equivalent If to the value of the services, the net effect is neutral. Some of the literature discussed in Section II of this study takes the view that equilibrium, taxes and public since taxes are viewed as in a state of long-run services are not capitalized the pricing mechanism for public services, and thus, the two do not affect housing prices. Since percentage property of taxes housing generally greater upon values, $1,500 of $90,000 house and another town. $10,000 lives Further suppose increase by The $90 increase their levied income as effect household with in a $250,000 that a is Suppose a discretionary income lives 10%, from $10 per $1,000 in value. generally lower-income households. household with income of are in a a discretionary house in property taxes the same in the $1,000 in value to town $11 per The family with the $90,000 home sees its represents 6.0% 10 of its discretionary income. The family with the $250,000 home rise from $2,500 to $2,750. only 2.5% of its The $250 increase represents discretionary income. increase has a more significant sees its taxes Clearly, the tax impact on the lifestyle of the household with less discretionary income. To the extent that community, higher households seek a more homogeneous property taxes become a price into higher income neighborhoods. who have more discretionary high taxes of entry Higher income households income are less than lower income households. sensitive to Accordingly, by choosing to provide a greater amount of public services and charging income lower higher taxes towns prevent to finance entry income households those services, into who higher their neighborhoods cannot meet the higher by tax price of entry into the community Similarly, by imposing restrictive zoning ordinances, such as requiring minimum lot or structure sizes for homes, towns promote homogeneity. relatively less minimum lot drives Even suburban expensive, the requirement of size or constructing a up the if cost of buying a land is acquiring a minimum size structure home and prevents lower income buyers from moving into the community. Thus, suburban higher taxes communities to other homebuyers. rates could and zoning are raise methods used their price of entry by for To the extent this occurs, higher tax be correlated with higher, rather than lower housing prices, contrary to such practices. expectations in the absence of Section of this II studies previous of study the for the its addresses the issue of observed and discusses public variables selected regression analysis, while Section limitations. of Section III provides further the reasoning behind the IV reports and the analysis, and discusses some interprets the results of of an overview between relationship services and property values. discussion of provides Section V summarizes the study, whether capitalization was actually policy some implications of the results observed. II) LITERATURE REVIEW The theory upon which most literature regarding property tax and public service capitalization is based was first developed by Charles M. Tiebout in his 1956 article, "A Pure Theory of Local Public Expenditures" [16]. suggested that individuals indicate their Tiebout preferences for local public goods by choosing to live in communities which meet their preferred provisions of local public goods. Public goods are partly defined consumed simultaneously by consumption subtracts as goods which can be individuals and no individual's from any other consumption (i.e., non-rivalrous consumption). example is is national defense. a problem of efficient At the which consumers indicate their The classic federal level, there allocation because market for private goods, there individual's unlike the is not a mechanism through preferences for public goods. If there of public were such a mechanism, goods would be adopted. Instead, an optimum amount provided and an inefficiencies appropriate tax occur because the government attempts to adjust its provision of public goods to what it perceives to be the desired level of the typical voter. To the extent that the desires of the typical voter (assuming such desires are properly identified) vary from what is socially optimal, the allocation is not efficient. Tiebout goods argued that (e.g., local efficient because consumers the allocation police and fire communities which preferences for local public services. in part because mobility far greater than among nations. who are dissatisfied with goods to more fixed, and conform to their Such efficiency can among communities is It is difficult for people the national provision of public pick up and move not as difficult public protection) is local budgets are relatively select be realized of local to another country. to move to another community But it is if they are dissatisfied with the provision of local public goods. The simplest form of the Tiebout model made the to the patterns are following assumptions: 1.) Consumers community are where fully mobile and move their preference best met. 2.) Consumers have full knowledge of differences among revenue and expenditure patterns. 3.) There is a large number of communities from which consumers may choose to live. 4.) Restrictions due to employment opportunities are not considered. 5.) There are no external economies between or diseconomies communities from pattern of community public services supplied. 6.) For every there is an optimum is defined services community size. in terms The optimum of the number for which the pre-set set, of residents collection of public goods can be provided at the lowest average cost. assumption Without introduces scarcity it, regardless communities of the into the could double This model. in limited availability size of some public goods such as parks or beachfront. 7.) Communities attract below new average costs at the the optimum residents in order size to of providing public optimum size try to seek lower goods. keep to the Those population constant. Given this set of consumers who goods in assumptions, Tiebout theorized that are discontent with the their communities will move to satisfy their preference patterns. to move, reveals goods. Thus, in provision of public the consumer's communities that Moving, or choosing not demand for the long-run, communities are local public made up of homogeneous populations with relatively uniform preferences for local public allocation of goods. public This results in goods than the a more efficient allocation at a national level because the consumers indicates their goods a in manner heterogeneous aggregate, population assuming a which conform of communities willingness to which cannot each community demands goods choice on be pay for national market level. by In for public only the number of to the public accomplished a national by a the goods, units of public preference patterns for that community. The total demand (the sum of the demands of the individual communities) more preferences of consumers closely represents than a central the true government's attempt to determine the typical consumer's preferences. Tiebout model. recognized a few of the limitations First, the costs of moving are a cost of indicating demand. To public the extent that goods rises, between remaining the costs of residents are in the providing local faced with community increased costs (which might and of moving, the less optimal is within the community. relocate indicate an The greater the the allocation Residents unwillingness public goods are forced to the be greater than the perceived public goods to the choice incurring benefit) or incurring the costs of moving. costs of the to of who wish to pay more for remain in the community because moving costs restrict mobility. Second, the assumption diseconomies ignores public goods the benefit provided example, if benefits (except to Town of no A by an has good the extent 15 external economies to one community adjacent law of the community. enforcement, that criminals or Town For B who might otherwise have chosen to commit crimes in Town A are driven to Town B because enforcement in of Town a perception B). The externalities occur suggests that integrating the communities, provision rather than of extent inferior to law which such there is some benefit to of these public having them goods provided among by each individual community. Third, although the model assumes there are a large number of jurisdictions from which the consumer can choose, it is clear that there is ultimately a limit to the number of communities which can be formed. obtain exactly the amount his preferences. Thus, found at national level or her inefficiencies Each consumer cannot of public goods the which reflects some of the same appear to a lesser extent in the local public goods model. Despite these limitations, Tiebout's model the point of reference for exploring consumers do pay provide a greater limited, however, more for housing amount of to the discussion of a relationship public goods and a serves as the possibility that in communities local public extent that goods. there is which It is not a between a preference for more price associated with obtaining them. The process described is merely one of choosing a community based upon its provision of public goods, without reference to the payments necessary payments could be made in the to obtain those goods. Such form of taxes or in terms of greater housing prices, but at its basic level, the Tiebout model does not raise this issue. Accordingly, we cannot conclude directly from homebuyers express the model a preference whether in for more public fact, goods by bidding up home prices in towns where more public goods are provided relative public goods. to prices The model explanation regarding the in towns which only serves provide fewer to provide an relatively homogeneous nature of the preference patterns within a community. Hamilton pointing [4], out that communities without a was critical of the mere was not public goods. He viewed the zoning restrictions, as the an urban area jurisdictions. with a to ration the therefore serve as model, choice among for efficient allocation consumption of property tax, in concert with pricing mechanism when used in large He argued further city, property taxes do not Tiebout availability of sufficient pricing mechanism the number of independent [5], that in the central act as a pricing mechanism and an excise tax which increases the cost of central city housing relative to suburban housing. Hamilton argued that the Tiebout model characterized a situation where the poor would follow the rich into suburbs as "free-riders," hoping base of the wealthier communities. would continue to form new would continue to follow. in low tax in these types of to benefit from the The While increased eliminating land users including the poor housing prices entry, vacant land communities would eventually be property wealthy, in turn, communities to which rate towns would discourage larger tax developed by all low-income price differentials due to housing, the favorable tax base. Restrictive zoning prevents this, incorporated allowing it into each his version of community to the Tiebout enact a residents to consume a minimum and Hamilton law model, requiring amount of housing. all He also introduced a proportional property tax to the model. In such a model, Hamilton hypothesized that in a state of long-run equilibrium, households would community whose minimum housing to the household's household in a requirement is less, it can to a town with community will long-run, there more have a are households consuming the requirement was just equal desired housing were located reside in consumption. community where If the the minimum increase its utility by moving restrictive zoning larger base. tax homogeneous their utility because Thus, communities that in the with all maximizing combination of housing and public goods. Under these conditions, property taxes serve purely as and do not according to Hamilton, a price for public services affect the price of housing. The property tax acts almost as efficiently as a head tax because the system of restrictive zoning communities goods. ordinances with similar Thus, the results in demands for homogeneous housing and public cost of goods can be spread evenly over households by charging a property tax which is almost uniform across the community. Each community has the number of a different tax based upon its housing consumption. demand for public goods and The tax is less efficient than a head tax only because the model calls for sorting by preferences for both housing consumption and public goods consumption. If a head tax were imposed, sorting would regard to the demand for occur only with public goods, leading to a more efficient benefits tax. Hamilton goes on occurs only in to point out that capitalization a state of disequilibrium where there is a scarcity of some type of desired community. land uses are shifted to activities which earn an economic rent (i.e., and a rent in overhead) attributes. property A values excess of market returns by providing positive or becomes the taxes are in negative dependent short supply, capitalized into higher prices. short supply, communities of profit desired community relationship upon communities which are in short supply. lower In that case, the with types of If communities with lower taxes will be If public services are in which provide relatively more public services will experience a capitalization of those services higher long-run into equilibrium, accordance however, with preferences the housing results housing in prices. sorting of consumption property taxes In communities and public in goods representing the price of public services and capitalization does not occur. Hamilton's conclusion from an empirical study [5], was that the Tiebout model operates where there choose so public are enough that property taxes goods. limited and school In the the property only at the suburban level districts are viewed as the central city, tax is viewed 19 from which choices as an to price of are more excise tax which increases the cost of housing and decreases consumption relative to the suburbs. This conclusion Ihlanfeldt [7], was whose further 1984 Metropolitan Statistical Areas in central city housing tax consumption agreement while payment of housing they 30 associated tax increase, with for local excise a greater Ihlanfeldt stated hypothesis that represent an Standard (SMSAs) indicated a decline services. as payment of by was either a neutral effect or was with Hamilton's are viewed study consumption with a while in the suburbs, there a higher substantiated property taxes services in tax his the suburbs on housing in the central city. The indication taxes and of a positive property values in Ihlanfeldt and explanation. Hamilton Those relationship the suburbs observed studies communities has a with relationship between taxes and property values perceived value of a present which exceeded the between benefits from present value of the a in the plausible positive could have public goods taxes required to receive those public goods. Hamilton's point would be that such an excess of benefits over costs was not perceived in the central city. Barriers to mobility such as restrictive zoning and the limited choices of jurisdictions within the central city serve to prevent central exercising a jurisdictional choice in accordance Tiebout model. Faced with those city consumers from with the barriers, central city consumers who view the tax as too expensive relative to the 20 benefits provided reflect that view by consuming fewer housing services rather than moving. In addition to the practices of restrictive zoning and higher taxes, another towns concerns public good. form the concept of which serves to acquisition can good. be However, space involves the acquisition is can convince land, the expense of residence, viewed as more the extent procurement of of others from accurately a such open enjoying it, described as a Further, if the town or Federal government to purchase townspeople might not of land To the enjoyment of a public good. taxpayers Such benefit of more open space, since the the state the cost for development. the exclusion collective good than the than a space within its community. increase taxes and remains available public suburban a collective rather that the townspeople enjoy the the used by A town can vote to acquire tracts of land for the preservation of open an action of barrier who, ever enjoy the due to open space at their benefit from the location the open of space. Groups of homeowners who suddenly become very excited about establishing motive of "conservation increasing the areas" price often of have entry the into true their communities. The view of value a perception of a surplus of benefits present services value from of the was advanced public services taxes required by Oates [11]. in the present relative to to pay In for the those describing the Tiebout model, Oates stated that the property tax liability represents the price of consuming local public services. consumer concerned the community surplus in with maximizing utility would in which he or she perceived the present value public goods relative to the tax payments required to of the future A seek out the greatest benefits from present value of the future receive those goods. Thus, all else equal (including tax rates), property values should be higher in the community providing the more attractive package of public goods. To test towns in level this hypothesis, Oates gathered northeastern New of output of expenditures per the use of public pupil. this Jersey. services, variable public services spending might not accurately of primary and secondary his proxy educational for the Oates used He recognized the noneducational quality As data from 53 because and because school limitations of it neglected variations in reflect differences services. education in the Nonetheless, represented the since largest single item in most public budgets and was of importance to families with proxy for children, he felt it was a reasonably good exploring the potential capitalization of local public services. Other variables utilized in the Oates model were: 1.) Linear distance of the municipality from midtown Manhattan as a measure of proximity to the major employment center 2.) Median number of rooms per dwelling as a measure of size. 3.) Percentage of homes in the community built since 1950 as a measure of of the the age housing stock. that this income; Oates reasoned 4.) Median family such as reasonable proxy for intangibles was a attractiveness of the would select tax rate, defined in all of presumably families wealthier stated that He the house. beauty of charm or neighborhood, or physical higher quality residences. 5.) Effective property nominal tax rate times the these studies as the assessment ratio used in the community. of families 6.) Percentage the study that values was with poorer weighted downward median to explain and in who tend to dwellings families homeowners would family income of rent, median by was reasoning here trying owner-occupied of communities be The under $3000. incomes with the community in a large renting population. After variables, regressing median Oates found taxes and that property values were and property values concluded on these negatively related and expenditures values were the home values of the that unaccompanied an by an positively related. coefficients increase increase in in in his property seven property per pupil Based upon equation, tax expenditures per he rates pupil would cause a reduction in rate increase were property values. accompanied by However, if a an increase in expenditures per pupil, the expenditure increase's positive effect might offset, or more effect of a higher tax rate. a validation of in offset the more to live in a it indicated that community which either relatively high-quality output of relation to the taxes negative This was seen as somewhat of the Tiebout model since people would pay provides a than charged, or public goods the same level of public goods at lower tax rates. Oates discussed the problem of using expenditures as a measure of the output of local public goods, a problem which exists in most of the studies conducted in this area. He also noted that the utilized in this represented points strictly true. Hamilton study) assumed observations of equilibrium, which was probably not In fact, [4], concluded it is that (as are the directly attributable point of use of cross-section data interesting to the results to his note that Oates obtained were observations occurring at a disequilibrium, and that at a point of long-run equilibrium, such capitalization would not have occurred. In response to a critique [13], Oates [12], including a all This than local public additional variable was property values, hypothesis. version of the regression variable for municipal spending functions other service. with ran another of his model by Pollakowski as was Also, whereas schools and debt positively related expected under Oates had per capita on the Tiebout concluded from his first model that capitalized rate property into differences were this values, partially model revised roughly indicated expenditures public all including tax complete capitalization of tax rate differences. communities near thirty-nine as a interpreted their separate The data further values. They strong positive and expenditures school student per variable. independent results as indicating a between relationship 4) and Environment; school expenditures Annual Transportation. remained Community 3) subdividing Public Health and Persons and Property; 2) Services; Social Columbus, Ohio, education into four categories: public services other than 1) Security of similar study of Ingene [8], conducted a Kohlepp and property suggested that spending on other services did not affect property values. While the and Ingene did find that services and theorized rather coefficients were not the signs for both public health variables were that than such outputs questioning the for services. a high relationship of local health and would negative. public level of services, security. and property again as a proxy of a crime problem rather Similarly, for decreased the negative public health values could social service problems in in to relationship for public safety between expenditures probably result They inputs validity of using expenditures The negative social services the security expenditures represented expenditures might be indicative than significant, Kohlepp and reflect public the community which demand for housing. However, one must keep in mind that these coefficients were not statistically significant in their study. The model also uses several developed for this study in the sample of the public budgets individual components towns to examine whether those particular expenditures have a significant on effect than Rather values. property dividing all expenditures into one of four categories, this at expenditures looks specifically study for per capita service and all budget items police, fire, education, debt other than those four. Two other studies addressed the metropolitan nature of the Tiebout process, that in commenting nonmetropolitan areas, capitalization of property taxes and public services was not as They theorized that in less urbanized more local, such that areas, employment opportunities were interjurisdictional mobility was limited. was more of housing because the supply distance were value elastic than of undeveloped of employment fewer constraints industry and of studied in North Carolina, excluding the one-hundred and six towns nine largest cities. Pasour [6], Hyman and significant. in the the ratio of the residential real urban Northeast land within opportunities upon entry Also, the supply was greater, into the was there construction value of land to estate commuting lower. the total In North Carolina in particular, most school funding was provided by the state, so there was relatively little variance in property tax rates across communities. Given these considerations, 26 Hyman and Pasour expected of public expenditure and property to find little evidence in were with line the services, dollars, on capita basis, a per for local taxes local of annual amount The was negative, Their proxy statistically significant. public North expectations. their effective property tax rate coefficient for but not of areas urbanized Using an Oates-type model, they obtained results Carolina. which less in capitalization tax and mildly was positive Hyman significant, but its magnitude (.21) was very small. and Pasour did not did their nor aggregate local tax examine any specific local expenditure, from assumes local expenditures in examine public study One aggregate. in to grants that state and collections equivalent are roughly expenditures that utilized proxy the the to local governments are roughly equivalent across communities. Carlson [10], McMillan and in Wisconsin, theorizing North Carolina sensitive property could caused rates in two-and-one-half times those in small cities there Pasour's similar results of would conclusions. property tax be Wisconsin rates in less be there. were Since nearly North Carolina, a study of test a good McMillan and capitalization was of Hyman Carlson and also concluded that public service to demand among communities to differences tax that low have similar study conducted a and experienced the Tiebout model less applicable to nonmetroplitan areas. Some of the reviewed here are: basic conclusions of the literature 1) The Tiebout model hypothesizes that consumers shop among communities and locate in the ones which best of local preferences for the provision meet their public goods. 2) In long-run equilibrium and property with zoning restrictions taxes, capitalization does not occur because taxes serve as the direct pricing mechanism for public services is more evident in 3) Such shopping among communities the suburbs relative to the central city. Property taxes are viewed in the suburbs as a cost of public services but in the central city as an excise tax. 4) Property tax and public expenditure capitalization is less evident in nonmetropolitan areas. 5) If capitalization does more with associated appears to it occur, education expenditures be than with expenditures on other public services. considered in The towns six miles from to twenty-one suburban areas of a would expect Kohlepp and of property Tiebout hypothesis in major metropolitan area. results similar to those Ingene, indicating a degree tax rate differences located from should provide Boston, and the operation of the evidence as to we this study are and Accordingly, of Oates and of capitalization differences education expenditures, if not all public expenditures. in III) THE MODEL To study the relationship between local public linear regression model is estimates by based upon data upon relies describing used. describing the median census data area property values, and residential expenditures in an value of housing the homeowners, & this study Tradesman [1], The data 1989. sold during actually homes Rather than relying upon by Banker supplied a multiple county Registry of Deeds for combines information from the each town describing the actual sales prices and financing terms of homes sold with the records of the local Property detailing the Assessors' offices This approach should provide homes sold. picture how of a more accurate with homes prices market the of the physical aspects varying characteristics than estimates of value could provide. Ten towns in the Greater Boston area were selected for stock, the from view order to In study. study looks at 1800 to housing sales of homes ranging in size square feet. 2200 expect homes in this size terms of the number of the model fairly homogeneous a attempts to prospective homebuyer general, one In would range to be fairly consistent in bedrooms and baths provided. simulate the situation faced choosing between Thus, by a reasonably similar homes in two or more towns which provide different packages of local public goods. The dependent the sale price variable in the regression of housing. A total of 336 equation is sales of homes ranging in size from 1800 to 2200 square feet was observed, excluding a true, "arm's length" sales prices those cases, In transactions. to be not appear which did few sales after the others and the were substantially lower than buyer and seller often had the same last name. this is Since property values at actual looks home sales to explain and attempts the independent variables, a number of subtract from the which add to or including expenditures, between the model public expenditures, and local terms of variation in the relationship study of a sales price (value) of the home. categories: Fiscal 1) the located; and 3) the of Physical aspects variables in three fall into variables independent The towns in general 2) homes sold; the homes which were Locational characteristics associated with the towns. The five physical aspect variables included age of the home at the time of sale, square feet of living area in the home, lot size the equation, these variables are In the number of baths. shown as AGE, SQ. FT., LOT, BDRMS and BTHS, respectively. The expected sign for AGE that people for an premium are willing to pay older house updating. range comprise be in a very portion of 30 need of be willing old house, homes in the thirty a larger It is assumed more for a new buyers might charm of expectation is that is negative. which might While some for the of bedrooms and in square feet, the number house than repair or to pay a the general to forty-year old the stock than homes modernized. been Such homes are not expected to command much of and systems electrical and heating of modernization that extent the to especially premium," a "charm have which century the of turn the at built replacement of the roof is required. expect larger homes to sell for a higher price. the are also are willing to the assumption that buyers positive, under Similarly, and BTHS BDRMS, LOT, signs for expected One would FT. is positive. The expected sign for SQ. pay more for more land, bedrooms and bathrooms. was fiscal data The Massachusetts by the provided Those Municipal Assistance Bureau. Department of Revenue, variables include the residential property tax rate (TXRT), expenditures police per capita (DEBT), debt service expenditures per public protection (FIRE), education expenditures per expenditures per capita capita (ED/CAP), fire (POL), capita per capita and all fire than police, other protection, education and debt service (OTH). The expected with theory the sign for TXRT is of negative in accordance tax capitalization. factors are held constant, If all other homebuyers would be expected to reduce the price they would be willing to pay for a home if faced with a higher property tax rate. of real property taxation An aspect is worth noting. portion of the in Massachusetts Massachusetts law allows towns to shift a residential property commercial and industrial properties. in individual towns by a vote tax burden to This is accomplished of the local town council to at up to 50% more tax commercial and industrial properties a decrease in offset by increase is of taxes the amount and those classified as levied upon residential properties suppose the total assessed value For example, open space. The properties. on those be levied would normally than of all properties in a town is $2.5 billion, and 75% of the total classified properties of comprised is value total value is space while 25% of the residential or open as from those classified as commercial or industrial. Suppose of shifting, the tax rate in in the absence further that in property tax revenues. generating $25 million are taxed at up to 150% industrial properties so that they industrial properties (1.5%), to $15 and increases tax the increases This shifting. in have been would rate the what absence the property value those revenue from properties to $9.375 million ($625 million times 1.5%). balance of residential open and accomplished at a properties. burden on properties. An must This of owners owner of paid $2000 in property shifting is residential a $200,000 can be (0.8%) on such tax rate of $8 per $1000 of the from raised be properties. space The result the million $15.625 only In $25 million, a property tax revenues at order to maintain of and commercial on per $1000 in total tax the The law to commercial and to shift the tax burden allows the town of property value (1.0%), be $10 per $1000 of the town would and home who a lower open tax space would have taxes without shifting instead pays $1600, a savings of $400. to implement shifting is the A factor in the decision percentage of has a If the town industrial properties. commercial and comprised of value which is total assessed very high percentage of residential property, there is very little to 150% of taxed at Even when shifting. gain by what the tax rate might have been without shifting, a small of amount and commercial cannot properties industrial generate enough revenue to provide significant property tax relief to a homeowners in a predominantly residential town. a is there when However, base large relatively of shifting can be very commercial and industrial properties, beneficial to homeowners. has a which large significant derive tax burden is by viewed the from opportunities If the shifting of businesses located on those properties. the might industrial base commercial and employment A town to consider. additional tradeoffs There are those as businesses costs relative significant increase in operating might be experienced in a nearby community a to what which can draw upon an essentially equivalent labor force, the firms might relocate. However, residential character does not towns which desire to maintain a might resort to shifting even if it order to discourage generate much revenue in commercial and industrial development in the town. A potential homebuyer will consider more than just the tax effect of the presence or commercial or potential absence of a large industrial property users. negative externalities base of Presumably, the associated with the presence of factories values. Thus, live in a town with a more order to If such dampen would this occurs, process have due to shifting in a town industrial base. commercial or decision-making incur "residential character" rather than to benefit from lower taxes a large does not make shifting worthwhile in a base to property to willing town which in a enough of with to decrease might be homebuyer a higher taxes potentially serve would a the property tax rate which is negative effect of an increased otherwise anticipated. POL. Based and Ingene study, upon the Kohlepp expect a negative sign and the police budget However, one could is associated with predict the sign difficult to It is one would hypothesize that an increase in a crime of an indicator problem. large per capita make an argument that expenditures for police protection indicate a commitment to therefore lead to increases in crime prevention and should values. property again This an input to public expenditures, which are measure of the and the of quality regarding the number of issue of raises the the using services, as a services. Statistics crimes reported per 1000 residents clearance rate (percentage of which at least one suspect reported crimes for was arrested) would be a better Unfortunately, the reporting of such statistics indicator. is voluntary in Massachusetts and one of the towns in the sample does not report. whether FIRE should have a Similarly, it is not clear negative or positive sign. Do large expenditures per Some towns have volunteer the relationship, such as wages? security associated is greater that there If one the town. contracts with unions negotiate argues full-time firefighters while others have fire departments whose factor affecting Might there be another fire prevention? commitment to or a with fires a problem capita indicate with a larger expenditure per capita full-time department, than a should be positively related with housing values. upon Based is (ED/PUP) per expenditures Magazine [ 14]). on should be pay more for Per pupil greater expenditures. for comparison Boston per pupil quality, people of school an indicator willing to better variable by supplied expenditures per capita or If education regarding (Data were pupil education are expenditures per and education positive. the reviewed, literature the for ED/CAP expected signs pupil of most housing in towns with expenditure is probably a less highly because it is correlated with INC/CAP and SAT than is ED/CAP. signs for The predict. amount greater bonds are expected debt service future tax per burden fully guaranteed sign in district) and residents will to the extent that would view are utilized that be a municipal The negative. bonds are backed by a separate a parking authority or user fees risking government. by the local to with a greater would be capita that circumstance However, to the extent that authority (such as are difficult OTH it seems that towns Intuitively, of DEBT and both particular water and sewer to cover service the debt, as being between debt is not as negative as be as strongly should not It the values in this circumstance service and housing clear. adding to relationship The burden. tax potential not users and the by paid for directly in the former case, but should only be positive to the extent that minority of a utilized by being residents who the fees. supporting it through the user bonds as by the good financed the public residents view are If a majority of the town's residents use the public good and pay the user fees, a benefits tax and should not the user fees are serving as be capitalized into housing values. spending included Categories of include health and welfare, and culture extent these public goods (though welfare are viewed However, Kohlepp expenditures that such and sign in spending are equation the categories. recreation and effect dependent of the general government and health and health and Since several different variable, its the upon have a culture and strong positive welfare expenditures 36 relative various expenditure example, if expenditures for For They in the negative included within this negative effects positive and health and expenditures indicated is the for OTH should values. property relationship with property values. kinds of To the and Ingene found a negative in a community, resulting social problems works, adding to as relationship between not significant) hypothesized and recreation. in a community, the sign "quality of life" be positive. public fixed costs, government, general OTH variable in the have a mild negative effect, other the variable will be Conversely, if the negative effect of health and positive. welfare for sign the outweighs expenditures components of variable, its the OTH effects of sign will be the positive negative. The locational characteristics of the a few They include the are indicators variables sample towns. of average combined SAT scores for high school students in the town (SAT) [14], (INC/CAP ) [17], the town's the distance Boston (DIST) [17], and the assessed value (% space income in miles (also capita from the town to percentage of the town's total which is classified as RES/OP) per residential or open by provided the Massachusetts Department of Revenue, Municipal Assistance Bureau). The use of average combined SAT scores to measure school quality. this. First, variables in the correlated be There are several problems with correlation matrix equation, with both ED/CAP were might when a SAT was INC/CAP and of a proxy for was run found income for the to be ED/CAP and highly correlated with each more is an attempt INC/CAP and other. per highly Thus, SAT capita and education expenditures per capita than for school quality. Second, between there capita in is a correlation a town and the income the percentage of students per who take the exam; the greater the per capita income, the higher the percentage of the per students taking capita income percentage of all is in students the exam. a town, whose Thus, the lower the smaller performance is the can be measured by the SAT, because a relatively larger percentage public schools, attend the these students system. school of students in these school an test, ranging Thus, a large number 54% to a high of 92%. from a low of towns sampled, took the the students SAT by the public the quality of the ten for Finally, on the so performance is unrelated to 76.8% of average of towns do not 13.8% of the students in these an average of Third, not sit for the SAT. of students in those towns do the exam systems do not take used in this model as a measure of school quality. Despite these shortcomings, it does seem reasonable to presume considering the that parents would look for some purchase of a home quantifiable measure of school quality and published reports of SAT scores might be influential in the homebuying decision. If potential for their children to the that score higher on the SAT is particular town, in a enhanced by attending the schools they are likely to be willing to pay that town. feel parents home in more for a Thus, the expected sign for SAT is positive. The expected sign for INC/CAP is positive. Oates used median household income in his model as an indicator of the stock, hypothesizing that wealthier quality of the housing buyers would demand higher more quality indicators quality homes. than the This model uses Oates model, upon the theory that as income rises, demand services increases, variable included. Household a income is for a more income but based for housing is still desirable measure, but most current available data for household income is for 38 towns in and two of the populations of 25,000 areas with the sample have populations below 25,000. land markets is that theory of urban as willing costs. commuting are costs commuting decreased Thus, with increased in accordance for housing to bid to their amount they are and will decrease the place of employment increases. live in close proximity pay more to People will land values decrease employment center the major distance from The basic is not clear. sign for DIST The expected capitalized into land prices. households with However, higher incomes who demand more housing are expected to move further out from the city and of high households at income costs commuting costs. of commuting are to benefit households the a tradeoff point land costs just offsets the where the benefit of decreased increased exchange for result of this process is a The ability to own more land. location costs in commuting accept greater points At too the for higher-income ability to purchase less great from the further out, expensive land. This concept becomes complicated in metropolitan areas such as Boston. the largest significant employment nodes Routes 128 and 495. of center employment Many of factor in While the Boston CBD is in the there area, surrounding Boston are on the residents of the towns in this sample could be employed in commute to another highly developed one town on a beltway and such that proximity to Boston the household's location decision. is not a Thus, it is difficult to predict the sign for the DIST variable. The predicted sign not clear due for %RES/OP is also to the factors discussed above with regard to tax shifting. If of avoidance potentially higher greater tax is one would taxes, between prices relationship potentially externalities more expect and %RES/OP. burden is more than important a positive If avoiding a important, the for the expected sign would be negative. Table variables in next page. 1 summarizes the expected the regression equation signs and is shown on the TABLE 1 - SUMMARY OF EXPECTED SIGNS EXPECTED SIGN VARIABLE AGE - SQ.FT. LOT BDRMS BTHS TXRT POL FIRE - ED/CAP ?7 ED/PUP + DEBT OTH + SAT ?7 INC/CAP DIST ?7 % RES/OP IV) RESULTS, INTERPRETATION AND DISCUSSION The results highest R2 coefficients with combining the of the regression equation the greatest and relatively number of little correlation significant among the coefficients are shown in Table 2 on the next page: TABLE 2 - REGRESSION RESULTS Value Coefficient t-statistic -154,050.000 3.036 -295.034 2.955 SQ.FT. -3.681 0.840 LOT +0.385 2.897 -523.957 0.134 BTHS +25,249.040 4.478 TXRT +46,420.810 7.501 POL +1,534.926 2.667 FIRE -1,357.020 3.709 +45.899 5.670 DIST -3,111.720 3.275 DEBT -1,446.950 4.817 -397.101 9.029 CONSTANT AGE BDRMS ED/PUPIL OTH * R2 = 0.498 * - Denotes coefficient significant not the at 95% confidence level The regression expectations, but results not all. 2 was the value of R2. were in One of the line with some overall surprises In most of the literature, the values for R2 were in the 70% to 90% range while the best obtained with this data was an equation which contained several only upon represent finer size may constrained physical The feet. One 2200 square 1800 to in size from homes ranging 51.8%. of the study is the focus possible explanation sales of an R2 of variables and had highly correlated pricing distinctions than can be explained by the variables quality of construction, architectural and neighborhood other such as size range, factors Within the equation. in the style, landscaping significant have could effects influence on prices. study most used more of Oates, physical variable and then of the use excludes income tax and and rate This study uses equation in the above significantly less somewhat correlated with expenditure per pupil. The elimination of correlated variable could explain this highly family to be highly correlated with because income data was found price utilized median account for housing quality, several physical variables to and the dwelling as his proxy for housing quality. income as his sold than example, used for of rooms per owner-occupied median number only regarding homes physical data the literature. used, this the variables difference in as a As far some of the reduction in the value of the R2 All of the variables in equation the were statistically significant at the 95% confidence level, with the exception these of SQ.FT. two variables definition of homes sample are 1800 is and BDRMS. The insignificance of also attributable to be sampled. When all to 2200 square feet, it to the narrow homes in the is not surprising that the significance of square footage is reduced relative the bedroom could an additional sampled, size range Also, within broader range of sizes. sample with a to a be throughout the house and could indicative of smaller rooms therefore have a negative effect on price. The regression results demonstrate regard with especially capitalization, expenditure a degree of public to police expenditures per capita and educational expenditures all for FIRE, DEBT, and OTH are However, the signs per pupil. Further, negative. does not there seem to be a demonstration of tax capitalization since the sign for TXRT is positive. An some provides the of examination correlation A explanation. the of variables matrix correlation [9] indicates that FIRE is highly negatively correlated (-.577) That is, expenditures per (-.514). with OTH and with TXRT are higher. tax rates and other expenditures are highly positively correlated The correlation of some and TXRT was found (.504). INC/CAP (.581). towns, be lower when protection services tend to capita for fire other (.665). other variables was also tested, highly correlated with INC/CAP to be is with each TXRT and OTH also highly correlated with price This leads to the hypothesis that in higher income prices and tax rates tend to be higher, as are expenditures for public goods other than education, police, fire, and homeowners debt pay service. higher In these taxes and greater share of those revenues towns, allocate a wealthier relatively to expenditures in the OTH Conversely, in lower income towns, category than to FIRE. lower, and so are expenditures in prices and tax rates are Less the OTH category. If this tax revenue to relatively greater share of and allocate a FIRE. wealthy homeowners pay lower taxes would help explain hypothesis is valid, it why the sign for TXRT is positive and FIRE is negative. an in income higher towns, associating their as are serving variables extent, these To a proxy for prices with higher taxes and lower prices with greater expenditures for Although the fire protection. is negative, sign for OTH and its effect could be masked its magnitude is not great, by the positive correlation with TXRT. A further possible explanation for the high positive relationship between TXRT and price lies in the correlation of total the percentage between which is assessed value classified as residential or open space (%RES/OP) and TXRT. The .742, indicating residential have and %RES/OP between TXRT correlation coefficient that towns which higher tax rates. are is predominantly This makes sense considering the ability of towns to shift the tax burden to commercial and industrial property. very little commercial and to gain by shifting Those towns which have industrial property have little while those with a large commercial/industrial base can derive a significant benefit in terms of reduced one would expect tax rates relatively However, residential property tax rates. on residential property greater in predominantly there can be negative externalities 45 residential Thus, to be towns. associated in a town to live greater taxes the avoid character and contribute to a incur prefer to would homeowners that some such base, larger commercial/industrial with a in towns with living more residential with a would This externalities. and between price positive relationship an indicator of the residential TXRT, with TXRT serving as nature of the town. which lends greater credence (.487), and per capita income buyers bid up Assuming higher income hypothesis. to this between %RES/OP also a positive correlation There is the prices in towns with more residential property, seeking commercial/industrial taxes. In it base, not is higher taxes a sense, the larger a to unreasonable to higher are less sensitive such homebuyers assume that with associated externalities avoid to as a can be viewed cost of avoiding externalities just as they are viewed as a cost of receiving public goods. sign for DEBT indicates The negative apparently potential risk indicator of of positive correlation an indicator that expenditures, cruisers, supported than services. The between DEBT and POL used to debt is including annual radio equipment, etc. capita as rather taxes increased user-fee service per debt view increased that homebuyers an strong (.838) could be finance some replacement The as a of capital police correlation between DEBT and TXRT is extremely low (.000322), which might be an indication that the towns in this sample do not incur much debt. 46 The negative to distance these from Boston suburban housing sign for DIST indicates in determining housing towns. values The decrease coefficient approximately additional mile away from Boston. of employment centers some importance in values in estimates $3,100 that for each Again, due to the number the area, it is difficult to determine what factors contribute to this outcome. With regard to the physical results also point bathroom. According bathroom adds since the importance to the over $25,000 the size makes sense of the that the greater importance. the out fixtures factors in the model, the in value addition of an Also, since the exceeds the cost of adding additional one additional equation, to a homes sampled and plumbing of an house. Again, was restricted, amenity would it be of cost of adding all of required for a bathroom far a bedroom, it is not surprising that the additional bathroom has a more significant impact on the overall value of the house. While AGE and LOT were statistically significant in determining value, the magnitude of the coefficients was so small that the overall effect on price was minimal. Overall, the housing value as effects of most significant variables indicated by the model TXRT and BTHS. As reasons to believe that the affecting were the positive discussed above, there are high positive value of TXRT is associated with some other factors such as the avoidance of negative externalities and a positive income elasticity for housing demand. These influences could have served to erase the expected negative effect of higher tax rates on explored the relationship between local public price. V) CONCLUSIONS This study expenditures and single-family property values in ten towns in Massachusetts. It utilized multiple to develop a hedonic pricing constant for linear regression model which attempted to hold several variables thought to affect the sale price of homes. The anticipated results were a negative relationship between prices and tax between prices and public goods expenditures. that other rates and a factors, such as externalities associated development, or positive relationship the desire to with commercial It appears avoid negative and zoning and taxation practices industrial designed to create more homogeneous communities with respect to income caused the relationship between tax rates and values to be expressed as public positive in the model. However, expenditures were negatively related with education expenditures This was Since in keeping with the towns capitalization. a large (an average of 45% of sampled here), strong make up this topic. portion of the budgets in this positive relationship indicator of price, positively related. previous studies of education expenditures local public budgets fairly per pupil were while some public is a expenditure suburban areas, in that sorting in themselves between prices which taxes and for preferences suburban homebuyers public goods. the price of finding of explained their This with partly because property taxes as viewed real people process of Tiebout the accordance goods operated public who concluded [5] and Ihlanfeldt, by Hamilton in studies to those found this study are similar The results of a positive relationship observed in was not the central city. the suburban versus study did not address While this central city issue, the phenomenon of property tax shifting in Massachusetts such a are allows for an additional explanation for It appears that homebuyers positive relationship. partly motivated at least towns with by a desire to locate in is characterized a residential character which by a scarcity of commercial and industrial development. Massachusetts, such relatively towns can tax high property commercial/industrial base from a part provision in the of property tax allows towns burden to shift commercial to and to locate be bid up there in spite of countering the tax rates small from benefitting Nonetheless, the desire tax burden, as having because their prevents them in such towns causes prices to relationship between rates the law which industrial properties. the greater be characterized In expected negative and property values. The tax rates and high correlation between income per capita, prices indicates that higher income buyers bid up housing in these predominantly residential towns and prices 49 are to the higher less sensitive They are taxes. willing to bear these taxes in exchange for preserving the residential nature of their towns. this hypothesis is that towns A policy implication of which consider the possibility of increasing their tax base by allowing more commercial and industrial development face property values will drop the possibility that residential due to increased and commercial increase the tax base by Thus, externalities. negative perceived industrial development enough to more than reduction in residential property the must offset this values to accomplish the goal of enhancing the overall tax base. Perhaps the addition of account for some of with the other variables which could the perceived externalities associated industrial development commercial and explanatory power of this model. Also, would enhance the use of median household income in place of income per capita might provide an even clearer relationship between income and the willingness to pay to avoid externalities. Finally, the use of all a broader database which includes single-family properties significance of footage. Such would probably the physical variables, a modification model's simulation prospective homebuyers. 50 the especially square would serve to of the behavior add to improve the of a greater array of APPENDIX I Sample Correlations Prepared by: Marc A. Louargand 51 Mon Jul 16 1990 Page 1 10:11:31 PM Sample Correlations --- ------------------------AGE VKC DAT .0165 -. 2844 .0618 .9828 336) ( 336) C336) 336) .2585 .7632 .0000 .0000 UBn OBS 1.0000 ( 336) .0000 .9828 336) .0000 ( ( -. 2844 336) .0000 PRC ( 1.0000 336) .0000 ( -. 2959 336) .0000 1.0000 -.2959 C336) C336) .0313 336) .5680 -. 0013 .0803 336) .1420 -. 2537 .0000 .0000 .0165 C336) .7632 C336) .9818 .0313 (336) .5680 LA .0667 C 336) .2229 .0803 (336) .0735 336) .1420 .1790 -. 0013 336) .9818 -. 2537 336) -. 0603 336) .0000 .2705 1. 0000 .0000 -. 0446 336) .4 154 C336) -. 0446 336) .4 154 1.0000 336) .0000 -. 0456 336) .4046 .0194 -. 0456 C336) .0194 .7225 ( .0618 336) .2585 ( .0667 336) .2229 .0735 336) .1790 -. 0603 336) .2705 336) .7225 336) .4046 -. 0065 336) .9061 -. 0267 336) .6261 .2150 .0230 -. 1361 -. 0190 336) C336) S336) .7289 AGE ( BDRMS ( -. 0585 336) .2847 -. 1276 336) .0193 ( FIRPC 1930 336) .0004 - .0423 336) .4398 EDPC - ( .2058 336) .0001 .0000 0566 .0224 -. 0155 .1907 C336) C336) C336) .336) S336) .0004 .9431 -. 0341 -. 2929 C336) C336) .0000 -. 0643 C 336) 6 -. .6828 .3012 .3856 336) .0000 -. 1534 336) .0048 .2882 0336) .0000 .7769 .5336 .2144 .0022 -. 0220 336) .6875 -. 1256 336) -. 0842 C336) .4668 336) .0000 .0686 336) .2094 -. 0720 336) .0069 336) .9000 -. 1667 .0445 1730 336) .0015 nte) plecie (sameff Co siz 52 .239 C336) -. 1097 -. 0039 .2317 336) .0000 -.0803 336) .1419 -. 0679 336) .0003 C336) -. S336) .0897 C336) -. 1953 .0125 .0033 336) .9513 .2133 C336) .0000 -. S336) 1.0000 C 336) .6741 S336) ( .0000 .0001 .2517 POI.PC C336) .0001 sie .1006 C336) .0213 .0761 .164 1 .0176 .7478 336) .1233 Mon Jul 16 10:11:31 PM 1990 Page 2 POPC ----FIRPC 1930 -. 0423 .0000 .0004 .4398 .2882 -. 1953 POLPC TXRT voo BDRS -. 0585 LS -. 0065 336) ( C 336) -.0566 -. 1534 -. 0267 ( .0224 .0230 -. 0190 ( TXRT .5336 .1006 .2144 -. 2929 .0033 336) .9513 336) ( .0000 0039 -. 0643 336) .2396 .0761 C336) C336) .2317 C336) .0000 C336) .0022 - .0220 336) .6875 -. 1256 336) .02 13 .0176 336) .7478 -. 0842 -. 0076 336) .8897 .1070 336) .04 99 -. 0332 336) .5438 .0970 336) .1233 .7289 1.0000 336) .0000 .0205 336) .7090 .0205 1.0000 336) C336) .7090 .0000 .0578 .1036 1.0000 -. 0059 -. 1164 .1415 .0578 .0000 .9139 .0329 .0292 336) .5933 .0640 S336) 0 -. 1372 C336) 336) .0804 C336) .1415 .1036 (336) C336) .1641 .0640 C336) .2419 -. 0687 C336) .2089 C336) C336) -. 0059 1.0000 .2419 .2089 .9139 .0000 -. 1599 336) .0033 -. 0076 -. 0332 -. 1164 1.0000 336) C336) -. 1599 C 336) .8897 .5438 .0329 .0033 .0000 .1070 .0970 336) .0759 C336) .0292 336) .0499 .2165 C336) .0188 S336) 336) .0001 .7311 ----------------- ---------- --------------- --------------- .0897 C336) ( --------------- C336) -. 0679 336) -. 0341 ( -.0697 FIRPC EDPC -. 1667 .336) 336) ( POLPC -. 0803 336) 336) .9431 .0804 ( .2133 .3856 .0004 -. C 336) BDRMS 336) ( .0125 ( .0003 .0001 .1907 336) ( .0000 .0000 .7769 1097 .0048 .6828 336) -. .0445 .1419 -. 1361 AGE C336) C336) C336) C336) ( .6741 C336) C336) -. 0155 336) -. C336) .0001 ( .2517 C336) C336) .3012 .6261 S336) DAT .0193 C336) 336) .2150 PRC -.1276 C 336) .2847 .9061 =A C336) (336) 336) .0759 -. 5141 .000 .336) 336) .0118 -. 1372 .5933 -. 5141 336) .0000 1.0000 336) .0000 .0288 336) .5983 .4721 336) .0000 .2368 336) .0118 336) .0000 -. 3273 S336) 0 .000 1990 Mon Jul 16 Page 3 10:11:31 PM a--------------------OBS ( TAN -. 2058 336) .0001 -. ( PRC .0015 .0000 .4668 .0499 336) .3623 ( .0686 -. 0417 336) S336) .2094 .0720 .5336 .6625 -. 1534 -. 2576 -. 0519 .0000 .0048 .0000 .3431 -. 1230 336) .0188 336) .7311 .0288 336) ( TINT C336) C336) .1544 .1243 1095 -. 2752 POLPC .2286 -. 0993 .0692 .0448 .0000 .0247 336) .6517 -. 0244 -. 0394 .6558 .47 12 .1949 .2116 336) .0001 .2862 336) .0000 .0306 336) .5768 -. 0687 .0259 336) .6359 .4721 .0003 336) .9953 .0000 - .3273 336) .0000 1.0000 336) .0000 ( 336) .5690 336) .0840 .3957 .5983 .2368 ( .0779 .6739 .4466 .0290 .0000 .0230 C336) .0842 336) .1234 .8247 336) ( .5807 C336) .0465 C336) -. 0312 ( .4746 C336) (336) .0000 336) BDRMS C336) .0000 .1191 .0001 C336) C336) .0242 336) .2165 C336) .0000 -. 0121 C336) DIST -. 0341 .0005 .0000 .9000 INCPC -. 2623 C336) .1880 ( 1879 336) C336) ( -. .0000 .0420 336) .4425 S336) EDPC ( C336) .0069 FIRPC .5931 S336) - SAT OTHPC .3376 - S336) .0000 DAT AGE 1730 336) 336) ( Iyns---c--SAT- DBTPC -. 3192 336) .0000 EDC C336) -. C336) C336) C336) .2093 .0003 -.1297 336) .0174 -. 0134 .336) .0575 .336) .2933 .0256 336) .6402 -. 1462 .0409 336) .4547 C336) .1513 -. 0693 .0055 336) .2048 336) .0073 .8062 .6884 .5036 336) S336) 0 .1011 .0000 .000 336) .0004 C336) .0641 -. 0214 336) .6955 -. 0716 336) .1903 -. 5773 -. 4060 -. 2389 .2415 -. 1923 336) .0000 .0000 C336) .1018 336) .0624 .9476 .9034 C336) (336) .0000 ------------------ ---------------------------------- -------------------- 54 C336) C336) .6647 336) .0000 .8376 336) .0000 336) S336) .0000 .0000 336) .0000 -. 0039 336) .9431 -. 1449 336) .0078 .5767 6) S33 .0000 .3573 6) S33 .0000 Mon Jul 16 1990 Page 4 10:11:31 PM ------- --------------------- ASVAL TOTEXPPC -. .3445 aBS 336) ( ( .0006 .0000 -. .4144 TWN 336) ( .0004 .7076 .1145 336) ( 336) ( .0000 .0358 .0627 .0603 DAT 336) ( 336) ( .2518 .2706 -. .0554 AGE 336) ( .0008 .0445 .1028 336) ( 336) ( .4165 .0597 .3641 .0571 LS 336) ( 336) ( .0000 .2964 BDRZ4S -. .0219 0975 336) ( 336) ( .6895 .0743 -. BA .3351 1283 336) ( 336) ( .0000 .0186 .2653 .7757 TXRT 336) ( 336) ( .0000 .0000 -. .1752 POLPC 336) ( -. ( .9637 -. 6375 336) ( 336) .0000 336) .5338 .6273 ( 1407 .0098 .0000 EDPC 0025 336) ( .0013 FIRPC 1827 336) ( .3115 LA 1928 336) ( .0000 PRC 1853 336) ( 336) .0000 55 Mon Jul 16 1990 10:11:31 PM Page 5 Sample Correlations ------------------------------------------------------------------------OBS TWN PRC DAT AGE LA DBTPC -. 3192 -. 3376 .0499 -. 0417 .2286 -. 0121 ( 336) 336) ( 336) 336) 336) 336) .0000 .0000 .3623 .4466 .0000 .8247 OTHPC .5931 ( 336) .0000 SAT ( INCPC ( .6625 ( -. 1534 -. 2623 -. 2576 336) 336) .0048 336) .5336 ( .3445 336) TOTEXPPC .0000 ASVAL -. 1853 ( 336) .0006 Coefficient ( .4144 ( -. 1928 336) ( (sample size) 336) .0358 ( .0000 -. 0993 336) .0692 .0247 336) .6517 .0779 336) -. 1095 336) -. 0244 .0840 -. 2752 336) .0000 336) .1243 336) .2706 .0627 ( 336) .2518 significance level 336) .6558 .0448 .0603 ( .7076 336) 336) .1234 .1544 .1145 336) .0000 .0004 336) 336) .6739 .4425 .1191 336) .0290 .0842 ( .0230 ( .0420 336) 336) .3957 .0000 -. 0519 336) .3431 ( .5807 ( .0000 -. 0341 ( .4746 336) .0000 336) .0000 DIST 336) .0242 .0000 -. 1879 336) .0005 .0465 -. 1230 336) ( -. 0394 ( 336) .4712 .0554 .1028 336) C 336) .3115 .0597 -. 1827 336) .0008 C 336) .4165 .0445 Mon Jul 16 DBTPC 1990 10:11:31 PM LS .0306 ( 336) .5768 OTHPC ( SAT -. 0687 336) .2093 Page 6 BDRMS .0259 ( 336) .6359 ( .1949 ( 336) .0003 INCPC .2116 ( DIST ( TOTEXPPC 336) .0001 .2862 336) .0000 .0571 ( ASVAL 336) .2964 ( .3641 ( 336) .0000 ( .5690 -. 1462 336) ( 336) .0174 .0073 -. 0134 .0409 336) .8062 336) 336) .2933 .0256 ( C 336) -. 1297 .0575 ( BA 336) .6402 -. 0975 336) .0743 TXRT .0003 -. 0312 .4547 ( .6647 C 336) .0000 ( -. 0693 C 336) .2048 ( -. 1283 .0186 336) .0000 .0004 .1011 ( 336) .0641 -. 0214 336) C 336) .0000 .6955 -. 1449 -. 0039 336) .9431 .3351 .2653 C 336) C 336) .0000 ( FIRPC -. 0716 336) .1903 -. 5773 336) .0000 -. 4060 336) .0000 -. 2389 336) .0000 336) .0078 .5767 336) .0000 .1752 336) .0013 -. 6375 336) .0000 -. 0025 -. 1407 336) .0098 ( .7757 336) .0000 336) .0000 336) .5036 .0219 .6895 -. 1923 ( .6884 ( .1513 336) .0055 C 336) 336) .9953 POLPC .8376 ( 336) .0000 336) .9637 Mon Jul 16 DBTPC 1990 10:11:31 PM EDPC .2415 ( 336) .0000 OTHPC ( SAT .1018 336) .0624 Page 7 DBTPC 1.0000 ( INCPC ( DIST ( TOTEXPPC ASVAL 336) .0000 336) .0000 .1761 .0000 336) .0012 .3573 -. 1603 336) .0000 336) .0032 .6273 336) .0000 .0063 336) 336) .0050 ( .2638 336) .0000 .0012 DIST -. 1603 ( -. 0645 ( 1.0000 336) .0000 .8928 336) .0000 ( ( .8928 336) .0000 1.0000 336) .0000 ( 336) ( .3177 336) .0000 .8077 336) .0000 ( .7123 336) .0000 .0000 ( 336) .2386 .0019 336) .7398 ( .5296 ( 336) .0000 ( .2386 336) .0032 -. 1686 ( -. 0182 ( INCPC .1761 336) 336) -. 1686 ( .0546 336) .3182 ( -. 0645 ( .9090 ( 336) .2638 336) .0000 .0050 336) SAT .1528 1.0000 ( 336) .9034 336) .0000 336) .0000 .1528 ( .5338 ( 336) .0000 -. 4093 ( .9476 ( OTHPC -. 4093 336) .0019 .3177 336) .0000 .2915 336) .0000 .2915 336) .0000 1.0000 336) .0000 .4049 336) .0000 .0188 336) .7311 .5956 .0315 336) C 336) .0000 .5651 Mon Jul 16 DBTPC 1990 10:11:31 PM Page 8 TOTEXPPC .0063 ( 336) ( .9090 OTHPC 336) .0000 SAT -.0182 ( 336) .5296 ( .0000 INCPC 336) .0000 DIST .5956 ( .0188 ( 336) .7311 TOTEXPPC 336) .0000 ASVAL 336) .0000 336) .5651 .2487 ( 336) .0000 .2427 ( 336) .0000 .0315 ( 1.0000 ( 336) .0000 .4049 ( 336) .7398 .7123 ( 336) .3182 .8077 ( ASVAL .0546 1.0000 ( 336) .0000 APPENDIX II List of Towns in Sample List of Towns in Sample Arlington Bedford Framingham Lexington Malden Natick Wakefield Waltham Wayland Woburn REFERENCES 1. Banker and Tradesman Real Estate Data Publishing, Banker and Tradesman 1989 Annual, Middlesex North and Middlesex South, Boston, MA, 1990. 2. Goodman, Allen C., "Hedonic Prices, Price Indices and Housing Markets," Journal of Urban Economics, Vol. 5, 1978, pp. 471-484. 3. Griliches, Zvi, ed. Price Indexes and Quality Change, Harvard University Press, Cambridge, MA, 1971, ch. 1. 4. Hamilton, Bruce W., "Zoning and Property Taxation in a System of Local Governments," Urban Studies, Vol. 12, 1975, pp. 205-211. 5. Hamilton, Bruce W., "Property Taxes and the Tiebout Hypothesis: Some Empirical Evidence," in Fiscal Zoning and Land Use Controls, Edwin S. Mills and Wallace E. Oates, eds., D.C. Heath and Company, Lexington, MA, 1975, ch. 2. 6. Hyman, David N. and E.C. PasourJr., "Real Property Taxes, Local Public Services, and Residential Property Values," Southern Economic Journal, Vol. 39, 1973, pp. 601-611. 7. Ihlanfeldt, Keith R., "Property Taxation and the Demand for Housing: An Econometric Analysis," Journal of Urban Economics, Vol. 16, 1984, pp.208-224. 8. Kohlepp, Daniel B. and Charles A. Ingene, "The Effect of Municipal Services and Local Taxes on Housing Values," AREUEA Journal, Vol. 7, 1979, pp. 318-343. 9. Louargand, Marc A., "Sample with data from this sample Appendix I. Correlations," prepared and included herein as 10. McMillan, Melville and Richard C. Carlson, "The Effects of Property Taxes and Local Public Services upon Residential Property Values in Small Wisconsin Cities," American Journal of Agricultural Economics, February, 1977, pp. 81-87. 11. Oates, Wallace E., "The Effects of Property Taxes and Local Public Spending on Property Values: An Empirical Study of Tax Capitalization and the Tiebout Hypothesis," Journal of Political Economy, Vol. 77, 1969, pp. 957-971. 12. Oates, Wallace E., "The Effects of Property Taxes and Local Public Spending on Property Values: A Reply and Yet Further Results," Journal of Political Economy, Vol. 81, 1973, pp. 1004-1008. 13. Pollakowski, Henry 0., "The Effects of Local Public Spending on Property Values: A Comment and Further Results," Journal of Political Economy, Vol. 81, 1973, pp. 994-1003. 14. Steinway, Susan, "How Your School System Stacks Up," (tables included within "Private Lives, Public Schools" by Margaret Pantridge) Boston Magazine, September, 1989, pp. 144-145 and 184. 15. Sullivan, Arthur M., Urban Economics, Richard D. Irwin, Inc., Homewood, IL and Boston, MA, 1990, ch.13. 16. Tiebout, Charles M., "A Pure Theory of Local Expenditures," Journal of Politcal Economy, Vol. 64, 1956, pp. 416-424. 17. Universal Publishing Co., Inc., Universal Atlas of Metropolitan Boston and Eastern Massachusetts, Stoughton, MA, 1988. (includes U.S. Dept. of Commerce, Bureau of Census data for 1985 income per capita of towns in sample).